Alibaba Cloud Model Studio:Recommended models

Qwen models

From most capable to most cost-effective — pick what fits your use case

Third-party models

Same API format as Qwen models

Understanding

Extract text descriptions or structured data from images and videos

Generation

Generate images and videos from text or images, with support for editing, reference, and high-resolution output

Text-to-speech

For audiobook reading, voice broadcasting, virtual avatars, and more

Music generation

Generate music from prompts or lyrics

Speech recognition

Dedicated ASR and LLM-based approaches — choose based on accuracy and flexibility

Speech-to-speech

End-to-end voice conversation without separate ASR and TTS calls

Integrates understanding and generation capabilities across text, image, audio, and video modalities

Convert text or multimodal content into vectors, combined with reranking to improve retrieval accuracy